ABNORMAL RETURNS
AT LOCKUP EXPIRATION
MASTER THESIS MAY 2018
An Emperical Assessment of Nordic IPOs
Authors:
Joakim Sylvest Helm (ID: 25873)
Peter Sommer Fabricius (ID: 47065)
Supervisor: Caspar Rose Hand-in date: 15 May 2018
Study programme:
M.Sc. in Applied Economics and Finance Physical page count: 97
Number of characters: 251,196 (~ 110.4 pages)
ABSTRACT
The purpose of this thesis is to examine whether abnormal stock returns occur when lockup agreements of Nordic IPOs expire. Furthermore, our objective is also to assess potential ex ante cross-sectional predictors that may impact such abnormal returns. We have structured our cross- sectional hypotheses according to three overall classifications of characteristics, namely 1) lockup characteristics, 2) IPO characteristics, and 3) lockup period characteristics. When performing an event study on a sample of 141 Nordic IPOs that have taken place from 2009 to 2017, we find significant evidence of abnormally negative price reactions at lockup expiration. Specifically, our event study shows that Nordic IPOs on average experience a significantly negative cumulative abnormal return of -1.1% surrounding lockup expiration. This bears the highly noteworthy and focal implication of invalidating the semi-strong form of the efficient market hypothesis. With this finding, we contribute to the existing field of research by (to our knowledge) being one of the first European studies to observe significant evidence of abnormally negative stock returns when lockup agreements expire. We furthermore provide a thorough analysis and discussion that discredits any exploitative opportunities of risky arbitrage, as the magnitude of the average cumulative abnormal return does not surpass concurrent transaction costs. In addition, when conducting a cross-sectional analysis of ex ante explanatory variables, we find evidence of three statistically and economically significant characteristics. Here, our results suggest that more negative abnormal returns will on average transpire when 1) a greater percentage of shares are subject to lockup relative to the overall free float, 2) stock prices show high levels of volatility leading up to lockup expiration, and 3) consecutively positive price ramps are observed prior to lockup expiration. We attribute our findings to the underlying implications of the downward-sloping demand curve and distortive effects that stem from information asymmetry and overconfident and contrarian investor beliefs.
1 TABLE OF CONTENTS
1 Introduction ... 6
1.1 Background and research question ... 7
1.2 Scope and delimitation ... 8
2 Theoretical framework ... 9
2.1 Efficient Market Hypothesis ... 9
2.2 Information asymmetry, signalling theory, and commitment theory ... 10
2.3 Downward-sloping demand curves and liquidity restrictions ... 11
3 Hypothesis development ... 12
3.1 Hypothesis 1: Abnormal returns at lockup expiration ... 13
3.2 Hypothesis 2A: Lockup duration ... 15
3.3 Hypothesis 2B: Percentage of shares locked up ... 17
3.4 Hypothesis 2C: PEVC-restrictive lockups ... 20
3.5 Hypothesis 3A: First-day returns ... 22
3.6 Hypothesis 3B: Underwriter reputation ... 24
3.7 Hypothesis 4A: Lockup period volatility ... 26
3.8 Hypothesis 4B: Early insider trading ... 28
3.9 Hypothesis 4C: Lockup period price ramp ... 30
4 Methodology ... 32
4.1 Research philosophy and approach ... 33
4.2 Sample selection and data collection ... 34
4.3 Event study approach ... 34
4.3.1 Defining an event and the event window ... 35
4.3.2 Estimation window ... 36
4.3.3 Market model and abnormal return ... 36
4.4 Expiration date analysis and variables ... 38
2
4.4.1 Abnormal volume ... 38
4.4.2 Transaction costs ... 39
4.5 Data selection for cross-sectional hypotheses ... 39
4.5.1 Independent variables on lockup characteristics ... 39
4.5.2 Independent variables on IPO characteristics ... 40
4.5.3 Independent variables on lockup period characteristics ... 41
4.6 Control variables ... 42
4.7 Regression methodology for cross-sectional analysis ... 43
4.7.1 Assumption 1 - The error term follows a normal distribution ... 44
4.7.2 Assumption 2 - No autocorrelation ... 45
4.7.3 Assumption 3 - Homoscedasticity ... 45
4.7.4 Assumption 4 - Model is correctly specified ... 46
4.7.5 Assumption 5 - Exogeneity ... 46
4.7.6 Assumption 6 - Linearity in the parameters ... 47
4.7.7 Assumption 7 - No perfect multicollinearity ... 47
4.8 Interview methodology ... 47
5 Event study analysis ... 48
5.1 Abnormal return at lockup expiration ... 48
5.2 Abnormal volume at lockup expiration ... 53
5.3 Transaction costs at lockup expiration ... 55
6 Cross-sectional analysis ... 58
6.1 Dataset inspection and descriptive statistics ... 58
6.2 Stepwise regression ... 59
6.2.1 Base model ... 60
6.2.2 Stepwise model 1 ... 73
6.2.3 Stepwise model 2 ... 75
6.2.4 Stepwise model selection by AIC ... 79
6.3 Final model ... 81
3
7 Implications ... 89
7.1 Implications for the company and the underwriter(s) ... 89
7.2 Implications for the rational investor ... 91
7.2.1 Arbitrageurs ... 91
7.2.2 Short-term investors ... 92
8 Conclusion ... 94
9 Limitations and future research ... 96
4 LIST OF FIGURES
Figure 1: Overview for categorization of hypotheses ... 13
Figure 2: Daily development of average abnormal returns (AAR) ... 49
Figure 3: Daily development of average abnormal volume (AAV)... 53
Figure 4: Transaction costs and absolute AAR within the event window ... 56
Figure 5: Scatterplot for CAR and DURATION ... 64
Figure 6: Scatterplot for CAR and UNDERWRITER_RANK ... 66
Figure 7: Scatterplot for CAR and PE_VC ... 68
Figure 8: Scatterplot for CAR and EARLY_INSIDER_TRADING ... 71
Figure 9: Scatterplot for CAR and SHARES_LOCKUP ... 82
Figure 10: Scatterplot for CAR and PRICE_RAMP... 85
Figure 11: Scatterplot for CAR and VOLATILITY ... 87
LIST OF TABLES Table 1: Daily development of AAR ... 50
Table 2: ACAR according to different event windows ... 51
Table 3: Daily development of AAV ... 54
Table 4: Transaction costs and ACAR for the event window (-2, +2) ... 56
Table 5: Transaction costs and AAR for the event window (-2, +2) ... 56
Table 6: Summary statistics for dependent and cross-sectional variables ... 58
Table 7: Multivariate regression output for Base Model ... 61
Table 8: Univariate regression output for Base Model ... 63
Table 9: Regression output for Stepwise Model 1 ... 74
Table 10: Regression output for Stepwise Model 2 ... 75
Table 11: AIC stepwise regression ANOVA table ... 80
Table 12: Regression output for Final Model ... 81
5 LIST OF EQUATIONS
Equation 1: Market model ... 37
Equation 2: Abnormal return (AR) ... 37
Equation 3: Average abnormal return (AAR) ... 37
Equation 4: Cumulative abnormal return (CAR) ... 37
Equation 5: Average cumulative abnormal return (ACAR) ... 38
Equation 6: Abnormal volume (AV) ... 38
Equation 7: Average abnormal volume (AAV) ... 38
Equation 8: Transaction cost ... 39
Equation 9: Base Model ... 60
Equation 10: Stepwise Model 1 ... 73
Equation 11: Stepwise Model 2 ... 75
Equation 12: Test for endogeneity (1) ... 76
Equation 13: Test for endogeneity (2) ... 76
Equation 14: Final Model (1) ... 79
Equation 15: AIC ... 79
Equation 16: Final Model (2) ... 81
6
1 Introduction
In November 2017, TCM Group A/S1 (TCM) launched an initial public offering (IPO) on the NASDAQ OMX Copenhagen. The IPO prospectus for TCM included a lockup agreement2 between the underwriters and selected existing shareholders of the issuing firm, where the existing shareholders contractually agree not to sell a certain percentage of their holdings for a pre-specified period after the IPO. This period is known as the lockup period, whereas the date on which the agreement expires is referred to as the lockup expiration date3. Once lockup agreements expire, pre-IPO shareholders (e.g. shareholding executives, other shareholding employees, private equity (PE) firms, venture capitalists (VC), institutional shareholders) are free to sell their shares to the public in accordance with insider trading regulations (Patel, 2018).
The IPO process of TCM was led by Carnegie Investment Bank and Danske Bank who functioned as Joint Global Coordinators. Sebastian Hougaard – managing director at Carnegie Investment Bank – emphasizes that a typical lockup agreement “is a way of communicating to the market that […] we are holding the selling shareholders of firms with an unproven track record to their promises from the IPO” (Hougaard, 2018, p. 3).
This is exemplified in the prospectus for TCM, which addresses two lockup agreements that restrict the selling shareholder (IK Investment Partners4) and shareholding managers and members of the board of directors for 180 and 360 calendar days, respectively. A management-restrictive lockup generally reassures the market that restricted executives will continue their efforts in accordance with firm’s best interests (Field and Hanka, 2001), whereas lockups for PE and VC firms indicate that they will neither attempt to cash-out immediately after the IPO nor in advance of impending bad news (Ofek and Richardson, 2000). Hence, as argued by Sebastian Hougaard, such lockup agreements “give some certainty to the market that we are not just dumping shares [and] that there actually will be someone who is left with part of the responsibility” (Hougaard, 2018, p. 1).
Although lockups are not a legal requirement for IPOs in the Nordic market, it has become common practice to restrict the holdings of pre-IPO shareholders for a pre-specified period, typically 180 or 360 calendar days (Goergen, Khurshed, and Renneboog, 2006). A key feature of lockup agreements is that the lockup period’s duration and related terms are stated in the IPO prospectus, thereby making the lockup expiration date a completely observable event. However, even though lockup-specific information is publicly available, previous empirical research suggests that insiders’ share disposals at lockup expiration tend to generate an
1 TCM Group A/S is a Denmark-based manufacturer and supplier of kitchen and bathroom furnishings.
2 “Lockup agreement”, “lockup provision”, “lockup restriction”, and “lockup period” are used interchangeably throughout the thesis.
3 “Lockup expiration”, “lockup expiration day”, “lockup expiration date”, and “expiration” are used interchangeably throughout the thesis.
4 The selling shareholder in the TCM IPO was Innovator International S.à r.l., which is owned by IK Small Cap I Fund (51%) and certain co-investors (49%) as limited partner entities. IK Small Cap I Fund is advised by IK Investment Partners Limited, which is a UK-based PE firm.
7 abnormally negative stock price reaction (Ofek and Richardson, 2000; Field and Hanka, 2001; Brav and Gompers, 2003). This contradicts the conventional outlook of the semi-strong form of the efficient market hypothesis (EMH), according to which investors are expected to have incorporated the lockup expiration event into their expectations at the time of IPO. To caution investors, it is often emphasized in IPO prospectuses that insiders’ share disposals may affect the stock price at lockup expiration. This can for example be observed in the IPO prospectus for TCM, which states that “after the expiration of such lockup obligations, these persons will be free to sell their shares. […] Any such sales of shares could have a material adverse effect on the public trading price of shares” (tcmgroup.dk, 2018, p. 49).
The lockup effect is therefore vital for TCM’s stakeholders to take into consideration. Since TCM’s 180-day lockup on IK Investment Partners is set to expire on 23 May 2018, a highly important question prevails on how the stock price reaction will transpire. This is where the purpose of our thesis has its outset, namely to examine whether lockup expirations influence abnormal stock price reactions for Nordic stocks, as well as consider which drivers are expected to have an impact on such abnormal stock price reactions from an ex ante perspective.
1.1 Background and research question
This thesis has its outset in a broad extent of previous research that has observed significantly negative abnormal returns for stock prices when lockup agreements expire. One of the pioneers within this academic area, Ofek and Richardson (2000), investigated IPOs in the US market and reported significant evidence of the prevailing phenomenon. Since then, other notable studies in the US (Field and Hanka, 2001; Bradley, Jordan, Roten, and Yi, 2001; Brav and Gompers, 2003; Brau, Carter, Cristophe, and Key, 2004) have also found evidence of abnormal returns at lockup expiration, which has fuelled a novel and heavily debated topic among empirical researchers. One of the first studies that shifted focus beyond the US was conducted by Espenlaub, Goergen and Khurshed (2001), who assessed IPOs that had taken place on the London Stock Exchange. Here, they also observe negative abnormal returns at lockup expiration, however, their findings do not bear any statistical significance. Accordingly, Goergen et al. (2006) expanded this view by investigating the German and French markets where they found similarly insignificant evidence. Overall, empirical studies on the European markets have not observed evidence that complies with the initial findings of the US studies. Hence, these suggest a possible absence of significantly negative abnormal returns at lockup expiration in the European markets.
Our thesis seeks to explain this conundrum in modern day finance literature, namely why abnormal returns transpire at lockup expiration in markets that are assumingly efficient. We contribute to the existing field of research by specifically addressing the Nordic markets, which to our knowledge have not yet been explored
8 by existing literature. Furthermore, we seek to investigate explanatory predictors of abnormal returns at lockup expiration by employing a holistic approach in explaining the observed differences in abnormal returns.
The purpose of this thesis is thereby formulated by the following research question:
Research Question:
Do abnormal stock returns prevail when IPO lockup agreements expire in the Nordic markets, and if so, why?
The research question is supported by the following sub-questions, to specify the scope of this thesis:
1) Using an event study approach, is it possible to observe statistically significant evidence of abnormal stock returns when the lockup agreements of Nordic IPOs expire?
2) Are there any ex ante cross-sectional differences that may explain the abnormal stock returns at lockup expiration? If so, how do these differences relate to lockup characteristics, IPO characteristics, and lockup period characteristics?
3) Based on our findings on abnormal stock returns and ex ante cross-sectional differences, which focal implications must be taken into consideration by relevant stakeholders?
1.2 Scope and delimitation
The main objective of this thesis is to investigate abnormal price reactions at lockup expiration, as well as examine a selection of ex ante cross-sectional predictors that may have an impact on such abnormal returns.
In this sense, our assessment of cross-sectional predictors will only evolve around public information that can be attained prior to the expiration date. In turn, this implies that explanatory factors that occur at the time of lockup expiration are deemed to be irrelevant according to the scope of this thesis. We will therefore disregard potential reactions that transpire beyond the expiration date, thereby rendering long-term effects irrelevant to the focal purpose of this thesis.
In accordance with our research objective, we will assert our core focus to abnormal returns at lockup expiration and potential ex ante predictors. However, to put our findings into perspective and draw upon vitally implicative inferences, we will complement our analysis by considering the trading activity and transaction costs that prevail concurrently with lockup expiration. We will not provide an in-depth assessment of these factors in the same manner as with abnormal returns, however, they have been deemed relevant for complementing our findings and thereby ensuring a comprehensive discussion of implications with which we can address our research objective and the core purpose of this thesis.
9 Naturally, we must briefly remark that we acknowledge that a broad-spanned selection of characteristics may have an impact on stock prices (and accordingly also abnormal returns), which inevitably cannot be exhaustively accounted for within the scope of our thesis. Nevertheless, we will trifurcate our selection of cross-sectional predictors to represent distinctly influential characteristics, which collectively are considered to yield the most optimal representation of reality as possible.
2 Theoretical framework
In the following theoretical framework, we will elaborate upon fundamental economic theory that is required to construct our hypotheses on abnormal returns at lockup expiration. We will firstly consider the EMH, from which our research question has its outset, and subsequently cover the focal concepts of signalling theory, commitment theory, and the downward-sloping demand curve, as well as their influence in the context of lockup agreements.
2.1 Efficient Market Hypothesis
The EMH asserts that a security is always valued at its fair price, as it reflects all available information in the market (Fama, 1970). It is thus considered to be impossible for investors to profit on over- or undervalued stocks, as the price epitomises a fair equilibrium. The EMH exists in three different forms: 1) weak form, where current prices reflect all historical information that is available to the public, 2) semi-strong form, in which prices are assumed to incorporate and adjust for newly published information, and 3) strong form, which states that prices reflect all available information, including information that is held by insiders. Since the expiration date of a lockup is disclosed in the IPO prospectus, the semi-strong form of the EMH will be applicable for our study.
While the EMH is supported by a large body of previous research5 (Roberts, 1959; Samuelson, 1965; Fama, 1970; Jensen, 1978), there also exists a great extent of dissension on the theory6 (Greenspan, 1998; Barber and Odean, 1999; Shleifer, 2000; Malkiel, 2003). Critics of the EMH point to market deficiencies, such as the stock market crash in 2007/2008, to emphasize that stock prices are inclined to deviate from their fair value. Theory on behavioural finance attributes market imperfections to cognitive biases (e.g. information bias, overconfidence, and overreaction) and reasoning errors among investors (Kahneman and Tversky, 1979).
5 As emphasized by Jensen (1978), “I believe there is no other proposition in economics which has more solid empirical evidence supporting it than the EMH” (Jensen, 1978, p. 1)
6 As famously stated by Alan Greenspan (1988), Chairman of the US Federal Reserve’s Board, “There is one important caveat to the notion that we live in a new economy, and that is human psychology […] which appears essentially immutable” (Greenspan, 1988, pp. 1-2)
10 Investors are thus susceptible to biases that result in mispricing of securities, for example leading up to a stock’s lockup expiration.
Recently, it has been attempted to combine the two distinctions in a new paradigm. Here, the concepts of the rational and behavioural view are reconciled with an adaptive market hypothesis (AMH) (Lo, 2004), in which it is assumed that investors are neither fully rational nor completely biased. Instead, the AMH asserts that investors make investments based on trial-and-error and best-guesses. The fundamental logic of the AMH borrows from evolutionary science, as the changing business environment is believed to make old strategies obsolete and thereby calls for an adaptive realization of new optimal strategies (Lo, 2004). In line with this attention to the changing business environment, it has recently been debated whether the prevalence of artificial intelligence substantiates the EMH’s applicability (Marwala, 2015). As financial markets are becoming more infused with computer traders that incorporate artificial intelligence, it is argued that the EMH will consequently improve in applicability.
Although there exists an immense ongoing debate on the EMH, one cannot completely disregard its important role in economic theory. The EMH is a vital theoretical concept for our study since a significant prevalence of abnormal returns would signify evidence against the theory.
2.2 Information asymmetry, signalling theory, and commitment theory
An important theoretical explanation for the importance of lockups is that they function as a signalling solution to an information asymmetry problem (Brav and Gompers, 2003; Krishnamurti and Thong, 2008).
The implications of information asymmetry were originally popularised by Akerlof (1970), who coined the
“lemons problem” by metaphorically relating financial markets to a market for used cars. In this example, sellers have full insight into the condition of a car, whereas individuals who intend to buy the car possess an informational disadvantage. Since a buyer cannot ascertain the true quality of the car, a price will be offered that lies below the car’s true value. Consequently, high-quality cars are withdrawn from the market, thereby resulting in a market consisting of low-quality cars (i.e. “bad lemons”).
The information asymmetry problem is often referred to in conjunction with IPOs since a firm’s original owners may have information that potential outside investors do not know of. Therefore, outside investors are subject to an informational disadvantage. To circumvent this disadvantage and the implications that are demonstrated by Akerlof’s “lemons problem” (1970), certain signalling solutions must be employed to convey information to outside investors regarding the firm’s true value (Leland and Pyle, 1977). According to signalling theory, “good companies” (i.e. those with a greater possibility of success) must always convey such signals to the market. For example, Leland and Pyle (1977) suggest that original owners of a firm should
11 possess a significant stake in connection with an IPO, to convey that their interests are in line with the firm’s best interests. If no such signals are taken into use, the asymmetric information problem will prevail and thereby cause adverse selection in the IPO market.
As an extension to the implications of signalling theory, the concept of commitment theory suggests that
“insiders cannot just put their money where their mouth is; rather, they must commit to keep it there if the signal is to be credible” (Brau, Lambson, and McQueen, 2005, p. 520), for example by not only owning an equity stake in the firm, but also committing to locking it up in the IPO aftermarket. At the outset of an IPO, it is therefore not only important to convey signals to outside investors, but also assure that such signals are credible. This is especially important for firms where the information asymmetry problem is prevalent. Such firms must therefore employ signalling and commitment devices to not only signal that their interests are in line with the firm’s best interests, but also to convey their commitment to continue to do so in the IPO aftermarket.
2.3 Downward-sloping demand curves and liquidity restrictions
Lockup agreements also bear a significant effect from a market mechanism point of view, as they can impact the equilibrium price that is derived from supply and demand (Scholes, 1972; Shleifer, 1986).
The fundamental idea of the downward-sloping demand curve is that there exists an inverse relationship between demanded quantity and price (Ofek and Richardson, 2000). There are several explanations to why demand curves possess a downward slope. Firstly, the law of “diminishing marginal utility” suggests that a consumer’s marginal utility decreases when consumption of a product increases, thereby making the consumer willing to pay a lower price (Beattie and LaFrance, 2006). Secondly, the “income effect” emphasises that consumers can buy more of a good when its price declines (ceteris paribus), thus causing an overall rise in real demand7 (Romer, 2012). Lastly, the “substitution effect” proposes that when a substitutable good becomes relatively less expensive (ceteris paribus), it will appear cheaper than similar goods and thereby experience an increase in demanded quantity (Romer, 2012).
When supply increases and causes a shift in the supply curve (ceteris paribus), the interaction between the supply and demand curve will yield a decrease in price due to the downwards slope of the demand curve. This mechanism is therefore vital at lockup expiration, as there occurs a supply shock when investors gain access to previously locked up shares (Field and Hanka, 2001). This concept is also referred to as the “float effect”
(Ofek and Richardson, 2000; Field and Hanka, 2001; Cao, Field, and Hanka, 2004).
7 In other words, one can buy more of a good with the same level of demand when its price declines
12 Lockups are a prime example of selling restrictions that prevent the ability of certain corporate insiders to sell their shares in the firm. Apart from lockups, selling restrictions are also often an integrate specification of executive stock or stock-option based compensation contracts (Kahl, Liu, and Longstaff, 2003). Due to the restrictive nature of lockups, certain insiders that are subject to a lockup will bear the cost of having a moderately illiquid and undiversifiable portfolio until lockup expiration. From the perspective of the firm, such selling restrictions are vital for retaining key employees and other shareholders that have a substantial influence on the firm’s strategic and financial outlook. This will also incentivize such shareholders to undertake actions that improve the medium- to long-term prospects of the firm, rather than simply cashing out in the immediate aftermarket. Furthermore, as mentioned with respect to information asymmetry, it is also deemed highly beneficial for the firm to retain such shareholders to convey credible signals to the market that a continued commitment to the firm’s performance will take place.
However, from a converse point of view, such selling restrictions may also impair the potential welfare of the restricted shareholders. Kahl et al. (2003) establish an important theoretical standpoint regarding this matter according to a mathematically derived model. Here it is hypothesized that when a large portion of shareholders’
wealth is restrained by lockup restrictions, these shareholders will possess an inadequate ability to counterbalance the risk exposure of their portfolio. Holding all else equal, this will induce their portfolio risk and may consequently lead to a higher probability of share disposals at lockup expiration that are motivated by diversification needs. Kahl et al. (2003) assert great importance to this notion, as such share disposals may surprise outside investors if they do not recognize this possibility and incorporate it in their expectations in due time ahead of lockup expiration.
3 Hypothesis development
In the following section, we set out to develop a selection of hypotheses that are focal for testing and discussing the implications of lockup expirations in the Nordic markets. Each hypothesis will be based on previous research (in terms of theoretical concepts and empirical findings), as well as remarks from practitioners and our own financial reasoning. The core focus of the hypotheses is the negative cumulative abnormal return (CAR) that previous research has observed surrounding the expiration date of lockup agreements (Ofek and Richardson, 2000; Field and Hanka, 2001; Brav and Gompers, 2003). In addition, we intend to assess cross- sectional differences and their expected relationship with CAR from an ex ante perspective prior to lockup expiration. As illustrated by Figure 1 below, our cross-sectional hypotheses are split into three categories that each represent a set of characteristics that may have an impact on abnormal returns at lockup expiration.
13 Hypothesis 1 addresses abnormal returns at lockup expiration whereas Hypotheses 2 through 4 represent cross- sectional characteristics that may have an impact on such abnormal returns. With respect to the cross-sectional characteristics, the hypothesis development will assert focus on 1) how a specific characteristic affects the likelihood of insiders’ share disposals at lockup expiration, and 2) why investors may over- or underestimate the degree of such disposals, thus yielding an abnormal price reaction.
3.1 Hypothesis 1: Abnormal returns at lockup expiration
According to the semi-strong form of the EMH, there should not occur any abnormal price reactions when a lockup expires, as such information is publicly available in the IPO prospectus. Market participants can attain a broad selection of lockup-specific information from the prospectus, such as the duration of the lockup period, the number of shares that are locked up, as well as a description of all shareholders whose holdings are subject to a lockup. Therefore, any price reaction at lockup expiration should already have been accounted for, either in the offer price or during the first day of trading (Ofek and Richardson, 2000). In addition to this logic, it is assumed that a firm’s stock price is based on unbiased expectations by rational investors (Bradley et al., 2001;
Brav and Gompers, 2003). Therefore, even though unexpected events may occur at the time of lockup expiration for a specific firm (which would yield an unexpected price reaction), the abnormal return at lockup expiration should on average be zero to reflect the unbiased expectations by rational market participants. Brau et al. (2004) stress that it must be reasonable to assume that rational investors are aware of any potential reactions due to supply shocks or information asymmetry and will incorporate such in their predictions. Hence, regardless of informational misalignments or market mechanisms, investors are assumed to always establish their expectations in an unbiased and rational manner, thereby stressing that “a persistent negative abnormal return should not exist in an efficient market” (Green, 2017, p. 2).
14 Nevertheless, empirical studies on abnormal returns at lockup expiration have proved evidence against the semi-strong form of the EMH. Ofek and Richardson (2000) conducted one of the earliest studies, where they investigated the abnormal returns at lockup expiration for 1,662 US firms that had been taken public in the period from 1996 to 1998. Their results showed a negative CAR that ranged between -1% and -3% (depending on the applied event window) with which they suggested “a new anomalous fact against market efficiency”
(Ofek and Richardson, 2000, Abstract).
Field and Hanka (2001) also invalidate the semi-strong form of the EMH in a lockup context. Based on a sample of 1,948 IPOs that had taken place in the US from 1988 to 1997, they observed that firms on average experienced a significant CAR of -1.5% when applying a 3-day event window surrounding the expiration date.
They also highlighted a noteworthy observation that 63% of the sample firms had a negative CAR during the event window. These results are directly comparable to those of Bradley et al. (2001), who used a sample of 2,529 IPOs in the US during the exact same period. Their analysis yielded a highly significant average CAR of -1.6% when applying a 5-day event window. Furthermore, when narrowing their event window down to solely focus on the expiration date, they observed a negative abnormal return of -0.7% with similar significance.
Additional empirical studies have considered US IPOs in the exact same period (Brav and Gompers, 2003;
Brau et al., 2004), where evidence once again indicated significantly negative CARs around lockup expiration.
These studies attribute the negative CAR to the downward-sloping demand curve and accredit the use of lockups as a means of resolving information asymmetries. Hence, it is suggested that investors make untimely predictions that are furthermore distorted by an informational misalignment. Brau et al. (2004) additionally consider the notion on biases among investors, as they argue that investors become increasingly nervous as the expiration date approaches. This contradicts the fundamental assumption that investors are completely rational with unbiased expectations.
Espenlaub et al. (2001) conducted one of the first studies on abnormal returns at lockup expiration outside of the US market, by assessing 188 IPOs that had taken place on the London Stock Exchange from 1992 to 1998.
Here, they found a negative CAR that on average ranged between -0.5% and -2.5% (depending on applied event windows and sub-samples), which is comparable to that of the US studies. However, the results for the UK IPOs did not bear any statistical significance. The authors emphasized that the IPOs in the UK were more complex and diverse than those in the US, as they do not conform to a generally standardized lockup period of 180 days (Espenlaub et al., 2001; Brau et al., 2004). More recently, the authors revisited their original study and performed an analysis on 233 IPOs that had taken place in the UK during the previously applied period from 1992 to 1998 (Espenlaub, Goergen, Khurshed, and Remenar, 2013). Here, a more rigorous analysis based
15 on characteristic lines8 was performed to control for risk. However, once again the results yielded limited evidence of abnormal returns at lockup expiration.
Ahmad (2007) also investigates the UK market, with a sample of 268 firms that have gone public on the London Stock Exchange from 1995 to 2006. Here, several analyses are performed on the overall sample and detailed sub-samples. Ahmad (2007) finds significantly negative CARs across all models when applying event windows of 21 and 41 days. However, when narrowing the event windows down to a more concise length of 3 and 5 days, there is not observed any significant results. When an event is easily determined (as it is with lockup agreements, since the event is set to the expiration date) it is generally recommended that one considers narrow windows to assure robust representability of abnormal returns (Armitage, 1995). Hence, the presence of abnormal returns at lockup expiration for UK IPOs is conclusively discredited.
The pervasiveness of abnormal returns has also been assessed for firms that have gone public in Germany and France (Goergen et al., 2006) 9. The results for these markets add to the findings of the UK studies, as they also confirm the absence of significantly abnormal returns at lockup expiration. The authors conclusively stress the importance of further investigating price reactions to lockup expirations in the European markets.
Overall, previous studies have found mixed empirical evidence of abnormal returns at the time of lockup expiration. The early studies on the US market observed significantly negative CARs, however, when expanding the outlook towards the European market, the findings have been proven to be insignificant. It is therefore highly relevant to consider whether one can support the semi-strong form of the EMH when assessing the Nordic markets. Thus, our main hypothesis will be as following:
Hypothesis 1: We expect zero cumulative abnormal return at lockup expiration
3.2 Hypothesis 2A: Lockup duration
When assessing the implications of lockups in general, it is of great relevance to integrate inferences from the
“portfolio diversification hypothesis” (Pagano, 1993; Pagano, Panetta, and Zingales, 1998; Bodnaruk, Kandel, Massa, and Simonov, 2008) and the “informative selling hypothesis” (Aggarwal, Krigman, and Womack, 2002; Chen, Chen, and Huang, 2012). These hypotheses can be used to explain the motivation of insiders’
share disposals at lockup expiration. The portfolio diversification hypothesis suggests that insiders sell their shares at lockup expiration due to portfolio diversification and liquidity needs. It is therefore argued that such share disposals do not convey any information about the current valuation of the firm. Alternatively, the
8 A “characteristic line” is applied in regression analyses to summarize a specific security’s systematic risk and rate of return. The concept is a performance assessment tool within modern portfolio theory (Markowitz, 1952, 1959).
9 138 and 268 IPOs that have taken place on the German Neuer Markt and the French Nouveau Marché, respectively, during the period from 1996 through 2000
16 informative selling hypothesis suggests that insiders will dispose of their shares at lockup expiration, as they are privy to private negative information regarding the firm’s true value or outlook. These hypotheses are useful when assessing the probability and potential magnitude of share disposals at lockup expiration.
Ahmad (2007) argues that lockup duration functions as a credible signalling device to represent firm quality.
With respect to this notion, insiders of high-quality firms who possess positive information regarding the firm’s future performance are assumed to be less likely to sell their shares at lockup expiration. Therefore, insiders of high-quality firms have greater propensity to accept longer lockup periods, as they intend to signal their positive inside information regarding the firm’s promising outlook (Ahmad, 2007). Consequently, the insiders of high-quality firms that in fact do sell at lockup expiration are believed to be motivated by portfolio diversification needs rather than possessing an informational advantage. Conversely, Haggard and Xi (2017) emphasise that insiders of low-quality firms may also choose to sell their shares at lockup expiration due to negative private information. Consequently, share disposals in such scenario will be motivated by both diversification needs and an informational misalignment, thus implying a greater potential magnitude of share disposals.
Due to the prevalence of information asymmetry, outside investors cannot determine the true value of the firm and will bear greater uncertainty in their predictions regarding insider sales on lockup expiration. Holding all else equal, when insiders possess negative information (which is proxied by shorter lockup durations), there is a greater probability that outside investors will underestimate insiders’ selling activity, which will translate into an abnormally negative price reaction (Courteau, 1995).
In line with this logic, if longer lockup duration is an adequate proxy for high-quality firms, one may assume that insiders are more inclined to hold on to their shares. Therefore, the magnitude of share disposals will be relatively smaller such that there exists a lower probability of exceeding investors’ expectations. Holding all else equal, abnormal returns will be absent at lockup expiration in such a scenario.
The empirical findings of Ahmad and Jelic (2014) support the view on lockup duration as a signal of firm quality, as they observe that firms with longer lockup periods demonstrate greater survival rates (i.e. an indication of firm quality) than those with shorter lockup periods. When comparing the abnormal returns at lockup expiration between firms with longer and shorter lockup periods, Ahmad (2007) found evidence of significantly negative CARs among firms with shorter lockups. These findings were based on a sample of 268 IPOs in the UK during the period from 1995 to 2006.
Boreiko and Lombardo (2013) also find evidence of the signalling hypothesis when analysing abnormal returns at lockup expiration for 174 Italian IPOs in the period from 1999 to 2009. When applying an event window of 5 days (-2, +2), they find that lockups with durations of 180 days or less have a significantly negative CAR, both during the event window and on the expiration date itself. However, they did not observe significant
17 CARs among firms with lockups that are longer than 180 days. This sub-sample approach thus supports the notion that a longer lockup period conveys a signal of high quality, which in turn implies an absence of abnormal returns at lockup expiration.
The relationship between lockup duration and abnormal returns at lockup expiration is a heavily debated topic within previous empirical research. We therefore find it highly relevant to assert such focus to this study. We discussed this view with Sebastian Hougaard by addressing the Swedish IPO market, which historically has experienced a greater investor appetite within the small-cap and mid-cap segments. Here, it was emphasized by example that smaller firms are often associated with greater uncertainty, which stems from their opaqueness and short-spanned track-record. Sebastian Hougaard explains that “the risk is definitely higher […] and this will not be a decent investment until they have proven themselves […] Therefore, you would want them to be locked up for a longer time” (Hougaard, 2018, p. 6). This complies with Ahmad (2007), who argues that firms that encompass a higher degree of uncertainty must convey credible signals to outside investors by employing longer lockups in order to alleviate misjudgements by outside investors that stem from information asymmetry.
In summation, we expect that longer lockups are used as signalling device by high-quality firms to solve uncertainty issues that are related to information asymmetry. We therefore construct the following hypothesis:
Hypothesis 2A: We expect that lockup duration has a positive relationship with cumulative abnormal return at lockup expiration
3.3 Hypothesis 2B: Percentage of shares locked up
It is often cautioned in a listing prospectus that a supply shock of shares can occur when a lockup expires, in the sense that insiders may choose to instantly dispose of their shares (Field and Hanka, 2001). This is also evident in the prospectus for TCM, as it states that “the market price of the shares may decline as a result of sales of shares in the market […] Any such sales of shares could have a material adverse effect on the public trading price of the shares” (tcmgroup.dk, 2018, p. 49).
Abnormal returns at lockup expiration can be explained by the logic of the downward-sloping demand curve (Ofek and Richardson, 2000). When a sudden positive shift in supply occurs, the stock price is expected to drop according to the new intersection of the supply curve and downward-sloping demand curve. Since the expiration of a lockup is 1) entirely known and observable and 2) bears great potential economic importance given existing literature on supply shocks (Scholes, 1972; Mikkelson and Partch, 1985; Holthausen, Leftwich, and Mayers, 1990), the market should immediately incorporate the expected price impact into the stock price at the time of IPO10. Therefore, regardless of the magnitude of the potential supply shock that could occur at
10 Discounted at the required return on the stock
18 lockup expiration, there should on average exist no significantly abnormal price reactions. However, since outside investors cannot ascertain whether insiders intend to dispose of their shares at lockup expiration, they must construct a probabilistic prediction of possible outcomes. It therefore becomes of great importance to assess the implications of locking up a large proportion of shares relative to the overall free float of the stock.
The amount of equity that is held by insiders at the time of IPO can serve as a signal of the firm’s quality (Leland and Pyle, 1977), in the sense that insiders of high-quality firms are more inclined to retain a greater number of shares. They will thus discern benefits of diversification in favour of benefitting from the promising outlook of owning shares in a high-quality firm. However, Leland and Pyle (1977) add to this notion that lockup periods are relatively short (in the sense that they most often are equal to or shorter than 360 days) when compared to the strategic time horizon of firms in general. They therefore argue that the cost of holding additional equity will not be paramount, thus rendering it easy for insiders of low-quality firms to imitate such large equity holdings to attain a similar signalling value. Since insiders of low-quality firms may potentially sell out at the first chance they get, which is motivated by holding negative private information rather than for diversification purposes, one would assume a greater probability of shares flooding the market at lockup expiration than otherwise expected (Leland and Pyle, 1977).
One can also explain the price reaction at lockup expiration with the occurrence of speculative price bubbles.
Hong, Scheinkman, and Xiong (2005) develop a mathematical model with which they argue that the probability of speculative price bubbles is high when the free float of tradable shares is limited. For example, this is the case for firms where a large proportion of shares are subject to a lockup relative to the free float. In this setting, one will assume that investors have heterogeneous beliefs, are influenced by overconfidence, and face short-selling constraints (Hong et al., 2005). The stock price will therefore exceed the firm’s fundamental value due to two reasons. Firstly, there is an “optimism effect” where the stock price is biased upwards by heterogeneous initial priors. When the beliefs of initial priors are heterogeneous, the stock price will only reflect the beliefs of those that are optimistic, as the conservative investors merely remain inactive due to short- selling constraints (Miller, 1977; Chen, Hong, and Stein, 2002). Secondly, there is a “resale option effect”
where investors are willing to pay a price that exceeds their own belief of true value, as they expect other buyers to be willing to pay a premium in the future (Harrison and Kreps, 1978; Scheinkman and Xiong, 2003).
In line with this logic, when a large proportion of equity is locked up, the beliefs of outside investors are more likely to yield a valuation that exceeds the firm’s true value (ceteris paribus). A bubble thus arises as investors anticipate the option to resell their shares to those that have even higher valuations of the stock. In this sense, it is argued that firms with a higher percentage of locked up shares (relative to free float) have a greater probability of being affected by speculative bubbles. When short-selling restrictions are relaxed at lockup expiration, the price bubble will likely burst since the firm’s true value becomes evident to not comply with
19 the speculated price. Holding all else equal, this unexpected value realization among investors will result in an abnormally negative price reaction.
Brav and Gompers (2003) conduct an empirical study on the relationship between locked up shares and abnormal return at lockup expiration. They based their study on 2,871 US IPOs that had taken place from 1988 to 1996. They construct two percentage-measures of locked up shares, namely locked up shares relative to total shares outstanding and relative to post-IPO insider shares. It is observed that both proxies have a significantly negative impact on CAR at the time of lockup expiration, thus suggesting that a greater fraction of shares locked up is associated with more negative price reactions. Nowak (2015) supports this finding, as he observes that firms with a lower free float experience a significantly negative CAR at lockup expiration.
This result is explained by the notion that firms with a lower free float relative to locked up shares may potentially have a greater number of shares brought to the market at lockup expiration. Hence, the results can be attributed to the downward-sloping demand curve and outside investors’ inaccurate predictions concerning the magnitude of insiders’ share disposals at lockup expiration.
Nowak (2015) splits the total sample into two sub-samples by dividing the observations according to the median percentage of free float. He observes a consistently negative CAR when applying broad event windows11, whereas the significant effect disappears when narrowing the event window down to 5 days or less. Furthermore, it is observed that the magnitude of free float shares does not have a significant impact when performing a cross-sectional regression on the entire sample. This finding is explained to stem from the proxy- measurement of free float relative to locked up shares, as it unavoidably is an imperfect ex ante proxy-estimate of the actual number of shares that will be brought to the market at lockup expiration (Nowak, 2015).
Brau et al. (2004) do however find a significantly negative relationship between locked up shares and abnormal return when performing a cross-sectional regression, which is substantiated by a sample of 2,120 US IPOs from 1988 to 1998. Their cross-sectional regression is applied to a 5-day event window (-4, 0), from which it is found that CAR on average decreases by 0.03 percentage points for each additional percentage point of locked up shares relative to total shares outstanding. This relationship is observed to be significant at the 5%
significance level and is concluded to support the theoretical view on asymmetric information and the downward-sloping demand curve. As more shares become subject to a lockup, insiders will have less ability to signal the true value of the firm to the market. Combined with a greater possibility of shares flooding the market, which yields the implications of the downward-sloping demand curve, this effect will induce uncertainty among investors and cause an abnormally negative market reaction to the lockup expiration.
11 Nowak (2015) observes a negative CAR of -8.95% for firms with a lower free float when applying a 21-day event window (-10, +10), which is significant at the 1% significance-level. When applying a broader event window of 32-days (-1, +30), he observes a negative CAR of -13.19%, which also is significant at the 1% significance-level.
20 Sebastian Hougaard emphasizes that when stocks have a large proportion of locked up shares relative to the free float, this will “suppress the market price [because investors] know there will be a fairly large sell-down [at some point after lockup expiration]” (Hougaard, 2018, p. 3). Subsequently, we then refer to the theoretical assumption that investors anticipate the lockup expiration ahead of time and will therefore already have incorporated it in their valuation of the stock, to which Sebastian Hougaard argues that “a part of it might be [investors] remembering that the lockup actually expires” (Hougaard, 2018, p. 4). Hence, it is possible that investors do not fully incorporate their expectations for the lockup expiration in due time. This argument relates to the behavioural logic of Kahneman and Tversky (1979), who suggest that investors may not be consistently rational. Sebastian Hougaard adds that “it’s a lot about psychology [and] people preparing themselves for future sell-downs after a lockup. […] It isn’t as rational [but] it’s also difficult to know whether [a sell-down actually will occur]” (Hougaard, 2018, p. 4).
Due to the abovementioned implications of the downward-sloping demand curve, information asymmetry, and investor beliefs, we assert great importance in assessing the impact that the number of locked up shares may have on the price reaction at lockup expiration. We therefore construct the following hypothesis:
Hypothesis 2B: We expect that the percentage of locked up shares relative to free float has a negative relationship with cumulative abnormal return at lockup expiration
3.4 Hypothesis 2C: PEVC-restrictive lockups
IPOs are generally considered to be a favourable exit strategy for private equity and venture capital (PEVC) firms (Fenn, Liang and Prowse, 1995). With respect to PE firms, their investment strategy asserts focus on relatively short-term financial gains, which constitutes a trademark inherited in their investment strategy and operational DNA (Fenn et al., 1995). This notion is also emphasized by Sebastian Hougaard, who argues that
“PE owners […] will sell at the first given instance they can. That’s fairly common and anticipated by that market as well” (Hougaard, 2018, p. 3). PE firms typically have an investment horizon of 10 years (Fenn et al., 1995), in which capital must be committed, called, and redistributed back to the limited partners (LPs).
This suggests that holding periods for individual investments are relatively short-spanned when compared to other types of block investments, such as direct investments by institutional investors. Therefore, exit opportunities such as IPOs are often utilized by PE firms to sell their shares and redistribute the returns of the investment back to the LPs (Brau, Francis, and Kohers, 2003).
With respect to VC firms, Espenlaub, Khurshed, and Mohamed (2009) study the exit behaviour of VC firms in the UK and find that IPOs are the most preferred exit channel followed by M&A. In accordance with this finding, Goergen et al. (2006) argue that VC firms prefer to exit from their investments at the earliest
21 opportunity after a portfolio firm has been taken public. This is due to the fact that an IPO constitutes the final stage of such VC firms’ investment horizon.
Thus, due to the shared short-term outlook of PEVC firms’ investment profile in the setting of an IPO, there should (ceteris paribus) exist a greater likelihood of shares flooding the market once the PEVC-shareholders are released from their lockups. In addition to the higher probability of share disposals when PEVC-restrictive lockups expire, the magnitude of the float effect may also be substantially larger. This is supported by the assumption that PEVC-shareholders that do not make a complete exit at IPO often retain a substantial post- IPO equity stake (Fenn et al., 1995). When considering such large equity holdings in conjunction with the short-spanned investment horizon of PEVC-shareholders, this will entail a higher probability of a potential float effect that encompasses a substantial magnitude of shares.
The notion that PEVC-restrictive lockups are associated with a potentially greater magnitude of share disposals at lockup expiration is empirically investigated by Bradley et al. (2001). Here, they examine differences in expiration date trading volume between lockups that restrict VC firms and lockups for other shareholders. For VC-restrictive lockups, trading volume is observed to peak one trading day after lockup expiration. The trading volume is found to peak at a level that is nearly 100% larger than the pre-expiration level, which remains approximately 30% larger during the following month. For lockups that do not restrict VC-shareholders, Bradley et al. (2001) observe a similar pattern, however, it is deemed to be immensely smaller. These results are consistent with the argument that PEVC-shareholders tend to liquidate their positions immediately after lockup expiration, which leads to a substantial supply shock.
Applying the theoretical concept of the downward-sloping demand curve, a large sell-down at lockup expiration should imply an outward shift of the supply curve with a corresponding lower equilibrium price.
However, the float effect should already be incorporated in the price at the time of IPO, due to the public availability and economic significance of lockup information. Regardless of the probability and magnitude of the potential supply shock, rational investors should on average anticipate the trading volume at lockup expiration. Hence, in order for the float effect to persist, the theory prescribes that there should exist specific characteristics for PEVC-restrictive lockups that induce added uncertainty into the construction of the average outside investor’s expectations.
In line with this logic, Bradley et al. (2001) conduct an empirical study to investigate the relationship between abnormal returns at lockup expiration and VC-restrictive lockups. They find that lockup expirations on average are associated with significantly negative abnormal returns and that such negative price reactions are concentrated among VC-restrictive lockups. When such VC-restrictive lockups expire, Bradley et al. (2001) observe an average return of -1.25% on the expiration date, as well as a CAR of -2.81% when applying a 5- day event window. Although the price reactions for lockups that do not restrict VC-shareholders also are found to be negative and statistically significant, the VC-restrictive lockups yield an immensely more negative price
22 reaction. Bradley et al. (2001) suggest that such negative price reactions that transpire when VC-restrictive lockups expire are the result of an ownership transfer from VC-shareholders to traditional equity investors, and that the observed abnormal price response is a float effect that occurs even though the expiration date is anticipated.
Theory on information asymmetry may arguably explain why the float effect prevails although it is fully anticipated. In such a setting, PEVC-shareholders possess an informational advantage due to outside investors’
unobtainability of the expertise required by PEVC firms (Arthurs, Nam, and Park, 2014). Hence, it is possible that a strong exit pressure in the IPO aftermarket and expectations of large returns incentivize PEVC- shareholders to utilize their informational advantage, thus elevating their own returns by expropriating wealth from outside investors. In such a scenario, PEVC-shareholders promote earnings management to induce stock prices prior to lockup expiration. PEVC-shareholders may employ a “pump-and-dump” strategy, whereby they hype a firm’s stock and underlying fundamentals during the lockup period to inflate demand, and thereafter take advantage of the market’s bullish sentiment by selling their shares at a premium at lockup expiration (Arthurs et al., 2014).
In conclusion, we acknowledge that PEVC-restrictive lockups are associated with substantial float effects and potential complicative implications that stem from information asymmetry between insiders and outside investors. We therefore construct the following hypothesis:
Hypothesis 2C: We expect that PEVC-restrictive lockups have a negative impact on cumulative abnormal return at lockup expiration
3.5 Hypothesis 3A: First-day returns
Aggarwal et al. (2002) develop a model with which it is argued that insiders strategically accept IPO underpricing to ensure that sufficient value realisation has taken place when disposing their shares at lockup expiration. This view is substantiated by the notion that a greater extent of underpricing induces the probability of a larger first-day return, which subsequently attracts attention to the stock and creates an information momentum. Such an information momentum could for example stem from induced interest by research analysts or the media.
From a market mechanism point of view, this will shift the demand curve for the stock outwards and yield a higher equilibrium price. Holding all else equal, the underpricing effect will thus allow insiders to attain a higher price when selling their shares at lockup expiration. This underpricing effect generates an opportunity cost to the firm in the form of forgone IPO proceeds (Ritter, 1991), which is traded off in favour of the benefits of information momentum. The fundamental theory of Aggarwal et al. (2002) and Ritter (1991) assumes that
23 informational disadvantages are to some extent depleted by underpricing, which is supported by Chemmanur (1993) who argues that underpricing generates an information momentum that encourages investors and research analysts to produce information regarding the firm. In this manner, the more information that is provided regarding a firm, the more probable it will be that a firm can depict its true quality to the market.
Investors can therefore establish predictions with a greater degree of certainty.
Tolia and Yip (2003) conceptualize the impact of an IPO’s first-day returns on the stock price reaction at lockup expiration. In line with the categorization of Krigman, Shaw, and Womack (1999), they define IPOs with first-day returns between 10% and 60% as “hot” IPOs, whereas those with first-day returns of 0% or below are defined as “cold” IPOs. Here, the fundamental idea is that insiders are less inclined to dispose of their shares immediately after lockup expiration when an IPO has been identified to be “hot”. Conversely, investors will predict that “cold” IPOs will bear greater propensity to continue to perform poorly, therefore implying that insiders will make a sell-down of shares after lockup expiration to cut their losses. These actions are thus considered to impact the stock price negatively at the time of lockup expiration.
However, the empirical findings of Tolia and Yip (2003) do not prove any evidence in favour of their hypothesised outcome. Their sample consists of 407 US IPOs from October 1999 to September 2000, where the observations are categorized into four groups according to the magnitude of first-day returns12. Their results indicate that all IPOs, regardless of categorization, on average experience abnormal returns of -1% on the date of lockup expiration. Furthermore, they find that the negative abnormal return is only significant for “hot”
IPOs, which is contrary to their hypothesized outcome. Tolia and Yip (2003) acknowledge that this effect can be attributed to profit-taking behaviour by investors who anticipate the forthcoming release of locked up shares.
When assessing the daily abnormal returns during an event window of 21 days (-10, +10), Tolia and Yip (2003) observe a high degree of variation in the magnitude and significance of the daily abnormal returns. Although they on average only find a significantly negative abnormal return for “hot” IPOs on the day of lockup expiration, they also observe significantly negative abnormal returns on alternative days in the event window across all categorizations. Therefore, they suggest that future research should investigate whether a significant relationship can be observed across an entire sample of IPOs, by performing a cross-sectional regression rather than analysing distinct sub-samples.
Krigman et al. (1999) also investigate IPO underpricing and argue that “first-day winners continue to be winners over the first year, and first-day dogs continue to be relative dogs” (Krigman et al., p. 1015). With a sample of 1,232 large-cap IPOs in the period from 1988 to 1995, they show that first-day returns predict the
12 Tolia and Yip (2003) categorise their observations as “Cold IPOs” (first-day return of 0% or less), “Cool IPOs” (first- day return between 0% and 10%), Hot IPOs (first-day return between 10% and 60%), and “Extra Hot IPOs” (first-day return above 60%).
24 direction of excess returns during the following year. This compares to the findings of Affleck-Graves, Hegde and Miller (1996) who study the link between first-day return and short-term aftermarket performance, where they observe that the risk-adjusted returns during the first three months of trading are in the same direction as the initial first-day return.
In line with the theoretical outlook of Aggarwal et al. (2002) and Tolia and Yip (2003) we will investigate whether the induced information momentum and increased demand that is generated by larger first-day returns will impact the abnormal return at lockup expiration negatively. We therefore construct the following hypothesis:
Hypothesis 3A: We expect that first-day returns have a positive relationship with cumulative abnormal return at lockup expiration
3.6 Hypothesis 3B: Underwriter reputation
Underwriter reputation13 is theorized to convey credible signals of a firm’s true value to less informed investors. Arthurs, Busenitz, Hoskinsson, and Johnson (2009) find that lockup duration acts as a substitute signal to prestigious underwriter backing and conclude that lockup duration is a more costly and inferior signalling effect, such that it is used in lieu of underwriter reputation.
Brav and Gompers (2003) concur that high-quality underwriters are less likely to engage in wealth expropriation as the underwriters will engage in more rigorous monitoring and not allow any opportunistic behaviour for the insiders to take advantage of the outside shareholders. Thus, firms with high-quality underwriters can arguably reduce the risk of unexpected share disposals by insiders and thereby abnormally negative price reactions at lockup expiration.
One can argue that underwriter quality is associated with less information asymmetry regarding the firm’s true value in the IPO aftermarket (Yung and Zender, 2010). In line with this logic, underwriter quality constitutes a certification effect, whereby firms that are certified by high-quality underwriters experience induced transparency and a reduction in the severity of asymmetric information.
Given the bounded rationality of outside investors and the lack of operating history to verify the credibility of the firm’s signals, underwriter quality is a safeguard that mitigates the outside investors’ uncertainty regarding the firm’s true value and prospects, as well as the forthcoming actions of insiders at lockup expiration.
However, underwriter quality may not signify credibility in the same degree as a lockup agreement, thus not rendering lockups obsolete (Arthurs et al., 2009). Credibility is based on the bonding cost of the signal, where
13 “Underwriter reputation” and “underwriter quality” are used interchangeably throughout this thesis. This is based on the assumption that underwriters’ reputation is directly linked to their quality.
25 it can be argued that lockups bear a higher bonding cost than underwriter quality. Whereas lockups coerce insiders to commit to keep their money where their mouth is, underwriter quality does not entail the same degree of contractual guarantee in the aftermarket. The outside investor can observe the reputation and merits of an underwriter, however, whether this reputation and expertise will translate into positive externalities for the company remains uncertain. Although underwriter quality is difficult to imitate in the sense that high- quality underwriters are assumed to only “pick the best horses”, the bonding costs can be less severe for the firm than for the underwriter (Arthurs et al., 2009). Thus, the credibility of the signalling of underwriter reputation will depend on the incentives of the underwriter and whether these work in favour of the insiders or outside investors.
In this matter, Carter and Manaster (1990) argue that reputable underwriters are highly incentivised to ensure the greatest degree of transparency and representability of the firm’s true value. This implies that the bonding costs and thus the credibility of the signal will be high. Reputable underwriters have developed their reputation for strong due diligence over time, which has further resulted in a reputation of strong reliability and legitimacy. If a reputable underwriter was to issue shares of a firm, which provides incorrect information, the revelation of such deception would not only harm the firm but also the underwriter’s reputation. In turn, this would impede the underwriter’s ability to attract investors for future issues, thus creating a strong incentive for the underwriter to ensure that the information concerning the firm is as accurate as possible. Consequently, the stronger the reputation of the underwriter, the more severe will the downside be for the underwriter’s reputation, thus ensuring a stronger bonding cost to be signalled to outside investors.
Contrarily to this theoretical outlook, empirical evidence of the relationship between underwriter reputation and abnormal returns at lockup expiration suggests a negative relationship. Brau et al. (2004) measure underwriter reputation according to the volume of issues for each underwriter in the year prior to the IPO relative to the total market volume in the same year. They base the measurement of reputation on the year prior to the IPO to acknowledge the possibility of varying underwriter reputation over the time span of the sample.
Although statistically insignificant, they report a negative and economically significant estimate for underwriter reputation when regressing against a 5-day CAR. Accordingly, Bradley et al. (2001) report that the largest price declines at lockup expiration occur for firms with higher underwriter quality. Field and Hanka (2001) measure underwriter quality as the underwriter’s percentage of market share in terms of dollar values in the respective year of the IPO. They observe that IPOs with reputable underwriters experience somewhat more negative CARs with marginal significance. However, they also report that the significance of the underwriter quality effect is highly sensitive to the applied event window.
Dong, Michel, and Pandes (2011) analyse the relationship between the quality of underwriters and the long- run performance of IPOs. They define three underwriter functions that may improve firms’ long-run performance, namely 1) marketing, 2) certification and screening, and 3) information production. This